Could this be good for chess?

[Cardoso]

http://www.anandtech.com/show/3768/seam ... onsumption

Best regards,
Alvaro

[Michael Sherwin]

For testing changes to the source code of an engine by playing thousands of games quickly it would be fantastic. But, for distributed programming it depends on how fast the inter communication is between processors. If you have hundreds of processors setting idle while waiting for the main processor to get around to retrieving their results it could be far less powerful that it seems it should be. I get the impression that "good enough" means good enough for one processor to serve one internet connection and that means that anything else might be sub par when it comes to speed. And of course the atom is only 32 bits so a bitboard engine would already be at a disadvantage on this hardware.